Understanding the characteristics and long-term behavior of water-conducting fractures is important for the safety assessment of deep underground facilities in the plutons of an orogenic belt. This study describes the features of fractures and their fillings along with the long-term behavior of their hydrogeological structures in an underground environment based on the results of a geological investigation conducted at an underground facility in northern Kyushu. Fractures were classified into five groups on the basis of fracture orientation：A, B, C, D, and low-angle groups. The fractures in all groups are filled with epidote and quartz, which are formed at high temperature：from about 300 to 350 ^oC. Thus, the genesis of all fractures is the cooling process of granodiorite pluton. Almost all of the water-conducting fractures are included in the B group. Because a number of fracture fillings in the B group are filled by prehnite and crushed fragments of epidote and quartz, the fractures in this group alternated sealing and re-opening. The fracture characteristics in the B group are follow as：accompanying many alteration halos, long trace length, and cutting off other fractures. These results indicate that fractures in the B group have possibly functioned as pathways for groundwater flow in the long term.

This paper aims to develop a methodology for understanding the fault activity by observing exposed fault planes without covering younger strata. Based on purpose, faults developed in relatively homogeneous rocks such granitic types are investigated as follows; Gosuke Dam upstream outcrop of Gosukebashi Fault and Funasaka-nishi outcrop of Rokkou Fault were selected for the study of an active fault; and K-3 outcrop of Rokkou Houraikyo Fault was chosen for a non-active fault. The fault planes were examined under the three categories of continuity, cutting of older texture of the fault zone, and smoothness. These three categories were considered according to the following five measurements, with the third being subdivided into three ：（1）continuity ratio; （2）cut ratio of the older existing texture/strata; （3）planarity ratio; （4）roughness/undulation measurement; and （5）arithmetic average roughness obtained from photographic analysis. Of the five measurements, the first three have the potential to discern the threshold between active and non-active faults. Further examination and discussion are required to clarify their applicability as parameters for fault activity evaluation.